基于背书特征的电力行业联盟链账本篡改攻击检测方法  

Endorsement-Based Detection Method for Ledger Tampering Attack in Power Industry Consortium Blockchain

作　　者：王栋 杨珂 李达 周启惠[4] 张艳秋 阮倩昀 王瑜[4] WANG Dong;YANG Ke;LI Da;ZHOU Qihui;ZHANG Yanqiu;RUAN Qianyun;WANG Yu(State Grid Digital Technology Holding Co.,Ltd.,Beijing 100053,China;State Grid Blockchain Technology(Beijing)Co.,Ltd.,Beijing 100053,China;State Grid Blockchain Application Technology Laboratory,Beijing 100053,China;Institute of Information Engineering,Chinese Academy of Sciences,Beijing 100093,China)

机构地区：[1]国网数字科技控股有限公司,北京市100053 [2]国网区块链科技(北京)有限公司,北京市100053 [3]国网区块链应用技术实验室,北京市100053 [4]中国科学院信息工程研究所,北京市100093

出　　处：《电力建设》2023年第11期23-32,共10页Electric Power Construction

基　　金：国家重点研发计划项目(2022YFB2703401)。

摘　　要：针对目前电力行业联盟链缺乏高效账本篡改攻击在线检测方案问题,提出了一种基于背书特征的电力行业联盟链账本篡改攻击检测方法。首先,在电力联盟链绿电交易仿真环境中,提出并实现了账本篡改攻击。在此基础上,收集并提取了链运行数据中与攻击有关的背书特征,以构建起检测所需的数据集。最后,采用基于Boosting随机森林算法进行检测模型训练,并将模型非侵入式部署在电力联盟链上在线检测账本篡改攻击行为。测试结果表明,相比于基于规则的检测方法,所提方法对电力联盟链的运行负担较小,在识别耗时和区块链性能损耗方面都表现较好,仅造成4.03%的性能负担。与其他基于机器学习的检测方法相比,该方法可适配于多种共识算法,并具备较高的准确率,达到了95.75%。Addressing the lack of efficient online detection schemes for ledger tampering attacks in the current power industry consortium blockchain,we propose a ledger tampering attack detection method based on endorsement features.First,an attack on the state data of specific nodes in a power industry consortium blockchain was proposed and implemented in a green power-trading simulation environment.Accordingly,endorsement features related to the attack were collected and extracted from the chain-operation data to construct the required dataset for detection.Finally,the boosting random forest algorithm was used to train the detection model,and the model was noninvasively deployed on the blockchain for online detection of ledger tampering attacks.The test results indicate that the proposed method has a smaller operating burden on the power consortium blockchain than rule-based detection methods and excels in terms of identification time and blockchain performance loss,incurring only a 4.03%performance burden.Compared with other machine learning-based detection methods,this method can be adapted to multiple consensus algorithms and has a high accuracy of 95.75%.

关 键 词：电力联盟链 账本篡改 背书特征 数据集构建 攻击检测 

分 类 号：TM39[电气工程—电机]